AI - are we losing our brain capacity for independent research?

I think you're worrying too much. It is just a tool 

When you sayy research, what do most people actually do? They ask someone, who likely has limited knowledge, is unaware of what they don't know and maybe give a partial, biased or wrong answer. Or they look online at social media or even influence. Some may read newspapers or books, but these may be incorrect. Even source data and scientific papers have errors.

There is no perfect source. Something that consolidates information and tries to discount some of the most distorted is not a bad source.

People are happy with Wikipedia, which is not perfect, and originally crowd sourced information. AI is this just much bigger.

Is it perfect, no. But nothing is. It is getting better, yes. Is it better than asking a drunk guy down the pub, yes.

I think there are two real risks:

1. It is not the tool, or the information, but the loss of critical thinking. But this was never that strong and has been weakened by many factors, and was occuring before AI.
2. Not understanding the information it provides because it comes faster and denser than you can really process. Leading to an overestimation of reliable ability.

In time it will settle I think, find a place and limitations. 

If people took themselves out of the loop and abdicated all responsibility for everything and just let it run n the world, what would happen? Would government be better or worse? For the majority of things I suspect it would better. But we gloss over mutually incompatible and illogical things in society. If some of these are exposed it would not help. Life is grey, a machine is less good at balancing these, or perhaps shows us what we don't want to see.

Some very valid points. Yes, the loss of critical thinking which is in any case rarely taught in schools so you have to learn this yourself over decades of experience. Also, processing the sheer quantity of data available then assessing its value - yes, I can see the value of AI in processing hard data - but not so sure about leaving AI to make conclusions, especially where human social systems are concerned. Currently, many systems are highly flawed [I worked in the NHS and had experience of its unworkable investigatory systems, also the flawed legal system], because those setting them up never seem to look at many facets or test their system end to end, as engineers and scientists do.

And your final point is very interesting. For example, in our legal system, it is widely acknowledged that many innocent people are wrongly convicted but no one looks at  the whole criminal system, with a few to radical changes. Simply by commissioning a very expensive legal professional, some 'win' unfairly whilst others are not given justice.

Some very valid points. Yes, the loss of critical thinking which is in any case rarely taught in schools so you have to learn this yourself over decades of experience. Also, processing the sheer quantity of data available then assessing its value - yes, I can see the value of AI in processing hard data - but not so sure about leaving AI to make conclusions, especially where human social systems are concerned. Currently, many systems are highly flawed [I worked in the NHS and had experience of its unworkable investigatory systems, also the flawed legal system], because those setting them up never seem to look at many facets or test their system end to end, as engineers and scientists do.

And your final point is very interesting. For example, in our legal system, it is widely acknowledged that many innocent people are wrongly convicted but no one looks at  the whole criminal system, with a few to radical changes. Simply by commissioning a very expensive legal professional, some 'win' unfairly whilst others are not given justice.